[
https://issues.apache.org/jira/browse/PHOENIX-1273?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14145211#comment-14145211
]
James Taylor commented on PHOENIX-1273:
---------------------------------------
bq. is it not 2^n ranges instead of 2, where n is the length of the prefix?
That could be another way of doing it, but I was thinking to just generate two
more that would cover all these. The LikeExpression still runs as a filter, so
it'd filter out rows that didn't match but were in one of the two ranges.
Depending on how long the prefix is, the 2^n solution might start degrading.
Need to add these overrides for serializing likeType in LikeExpression (unless
I missed them?):
{code}
@Override
public void readFields(DataInput input) throws IOException {
super.readFields(input);
init();
if (pattern == null) {
try {
likeType = LikeType.values()[WritableUtils.readVInt(input)];
} catch (EOFException e) {
likeType = LikeType.CASE_SENSITIVE;
}
}
}
@Override
public void write(DataOutput output) throws IOException {
super.write(output);
if (pattern == null) {
WritableUtils.writeVInt(output, likeType.ordinal());
}
}
{code}
Not absolutely essential, but an end2end test for ILIKE used in a WHERE clause
in particular would be good. Bonus points for a test that uses a non literal
pattern :-). Take a look at VariableLengthPKIT.testLikeOnColumn() if you want
to copy/paste something similar?
> Phoenix should provide ILIKE keyword
> ------------------------------------
>
> Key: PHOENIX-1273
> URL: https://issues.apache.org/jira/browse/PHOENIX-1273
> Project: Phoenix
> Issue Type: Bug
> Affects Versions: 4.2
> Reporter: Jean-Marc Spaggiari
> Assignee: Jean-Marc Spaggiari
> Priority: Minor
> Attachments: PHOENIX-1273-v1.patch, PHOENIX-1273-v2.patch,
> PHOENIX-1273-v3.patch, PHOENIX-1273.patch
>
>
> Phoenix provides LIKE keyword but not ILIKE which is a case insensitive LIKE.
> Goal of this patch is to add this keyword.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
